If Google really wants to make switching to Gemini easier, it needs these 5 features

While chatbots can be genuinely useful tools, they often require a fair amount of tweaking and customization if you really want to make the most of them. As you can imagine, this also makes it harder to switch platforms once you’ve become accustomed to the AI of your choice.

Recently, I wrote about my experience switching from ChatGPT to Claude. I talked about Claude’s own import tools, as well as the many extra manual steps I took to get it all right. I’ll admit that while the process wasn’t impossible, it was far from ideal.

Luckily, Google is hoping to make things a bit easier for those interested in switching to Gemini. We recently learned that Google’s upcoming import tool will let you import not only memories but also chats. With that in mind, and as someone who’s gone through this switching process before, here are five things Google has to get right with Gemini’s import tool to set itself apart from the competition.

Self-defined import parameters would go a long way

Based on everything we know about the Gemini import tool so far, it seems like the memory import function works very similarly to Claude. That means it gives you a generic prompt that you then ask your existing AI, which returns an info dump you paste back into the import tool.

That’s all well and good, but I’d love to see a more customized approach. Instead of a one-size-fits-all prompt, it would be nice if you could select certain types of memories and behavior ranges you’re looking to import.

Maybe you want basic behavioral rules, like “don’t use em dashes,” or a standing preference for realism and plausibility, but you don’t want memories tied to past projects that are no longer relevant. Likewise, you might want to leave behind instructions that were specifically designed to address quirks in your old chatbot that simply don’t exist in the new one.

For chat imports, it would also be helpful to define parameters based on date ranges, the extent of a chat, or whether it was tied to a project that has clearly ended.

Detailed chat summaries for imported conversations

I like the idea of chat imports, as I had to do this manually for Claude with mixed results. That said, I’m hoping it’s not just a massive dump that becomes hard to navigate. For those with a few dozen chats, this might not be a big deal, but I had far more ChatGPT chats than I could ever reasonably sort through manually.

What would really help is a pinned summary when you open an imported chat for the first time, covering what the chat contains, when it was created, when you last worked on it, and where you left off. That context would make it much easier to resume projects in a new engine and decide what’s actually worth keeping.

Import review process for memories and chats

In addition to chat summaries, a review process before the import finalizes would be very helpful. Once you hit import, a pop-up could show summaries of the memory data and brief breakdowns of the chat logs it’s preparing to bring over.

You’d be able to uncheck certain items and, ideally, edit memories to add or update information before committing. Once a user trimmed down their logs and memories, only the relevant ones would follow to the new platform.

Redundancy checks against what the AI already knows about you

Not everyone will be switching to Gemini from a blank slate. If you’ve used the free version or had a past subscription, the import process could end up duplicating data or carrying over information that’s no longer accurate.

I ran into this exact situation with Claude. I didn’t go in completely clean, as I spent about two weeks with the raw experience to see how it compared to ChatGPT out of the box.

After importing, I found conflicting information about my career and even my preferred name, because ChatGPT hadn’t been updated on either. Claude would occasionally toggle between calling me by my actual name and a nickname I’d originally given ChatGPT, and it got confused about my professional background as well.

Essentially, some of the rules it imported were things I’d added to ChatGPT years prior that simply weren’t true anymore.

A redundancy check that flags contradictions for manual review, rather than making automatic decisions, would go a long way. It doesn’t need to be sophisticated; just surfacing the conflicts and letting the user resolve them would save a lot of cleanup.

The ability to turn chat logs into actionable behavior data or guidance

Some of my old ChatGPT logs were still useful and relevant, but many weren’t worth importing in full. The problem is that even outdated chats often contain behavioral context worth preserving.

A smarter solution would be the ability to parse a chat log for relevant behavioral data and convert it into a memory, without importing the full conversation. That would let users carry over meaningful rules and context without flooding Gemini with hundreds of chats they’ll never open again.

How achievable are the features likely to be?

It’s very likely that Google’s import tool will be fairly basic and only a mild step up from what tools like Claude already offer. That said, none of what’s described above is impossible. Granted, some features would land messier than others, but nothing I’ve mentioned is totally unrealistic.

The contextual awareness required for redundancy checks would likely need manual oversight, and the tool you’re switching from limits how directly Google can pull data, requiring users to fetch certain elements themselves.

Even an imperfect version of these features would be a significant step up from building your own import protocols from scratch. Whether Google sees a deeper import tool as a meaningful draw for users switching from competing platforms is the real question, and we should have our answer when the tool launches.